CN108879691A - A kind of method and device that extensive continuous tide calculates - Google Patents
A kind of method and device that extensive continuous tide calculates Download PDFInfo
- Publication number
- CN108879691A CN108879691A CN201810646054.3A CN201810646054A CN108879691A CN 108879691 A CN108879691 A CN 108879691A CN 201810646054 A CN201810646054 A CN 201810646054A CN 108879691 A CN108879691 A CN 108879691A
- Authority
- CN
- China
- Prior art keywords
- parallel
- operating condition
- load level
- parallel thread
- resource
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
- H02J3/04—Circuit arrangements for ac mains or ac distribution networks for connecting networks of the same frequency but supplied from different sources
- H02J3/06—Controlling transfer of power between connected networks; Controlling sharing of load between connected networks
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
- H02J2203/20—Simulating, e g planning, reliability check, modelling or computer assisted design [CAD]
Abstract
The embodiment of the present invention provides a kind of method and device that extensive continuous tide calculates, the method includes：According to the corresponding parallel thread resource of each operating condition distribution that the work information being obtained ahead of time is in the work information, with the corresponding load level of each operating condition described in parallel computation；It is the corresponding default load level distribution first object parallel thread resource of each of each operating condition operating condition in the parallel thread resource, to preset load level described in parallel computation according to predetermined stressor；During calculating the default load level or the load level every time, obtain the fine granularity calculated, the second target parallel thread resources are distributed in the first object parallel thread resource or the parallel thread resource for identical fine-grained calculating process, to execute the corresponding parallel computation process of identical fine granularity.Described device executes the above method.Method and device provided in an embodiment of the present invention can be improved the computational efficiency of continuous tide calculating.
Description
Technical field
The present embodiments relate to technical field of power systems, and in particular to a kind of method that extensive continuous tide calculates
And device.
Background technique
With the access of new energy power supply and novel load and the continuous expansion of power grid scale, the Static Electro of electric system
Pressure stability analysis is faced with new challenges.
Existing continuous tide calculation method efficiency is lower and poor for applicability, and the analysis for being especially unable to satisfy large scale system needs
It asks.GPU Floating-point Computation ability with super strength, carrier as parallel computation due to its massive concurrent thread the characteristics of is extensive
Applied to every field acceleration calculate in.However, existing continuous tide, which calculates the prediction-correction methods used, to be had relatively again
Miscellaneous control logic is taken in heavy load level and is augmented the special processing such as parametrization, so that overall calculation is not appropriate for GPU
Unified instruction parallel architecture, cause computational efficiency low.
Therefore, drawbacks described above how is avoided, the computational efficiency that continuous tide calculates is improved, to promote large scale system electricity
The efficiency and accuracy of stability analysis are pressed, becoming need solve the problems, such as.
Summary of the invention
In view of the problems of the existing technology, the embodiment of the present invention provide a kind of method that extensive continuous tide calculates and
Device.
In a first aspect, the embodiment of the present invention provides a kind of method that extensive continuous tide calculates, the method includes：
Corresponding parallel thread resource is distributed for each operating condition in the work information according to the work information being obtained ahead of time,
With the corresponding load level of each operating condition described in parallel computation；
It is each of each operating condition operating condition in the parallel thread resource according to predetermined stressor
Corresponding default load level distributes first object parallel thread resource, to preset load level described in parallel computation；
During calculating the default load level or the load level every time, the fine granularity of calculating is obtained, is
Identical fine-grained calculating process distributes the second mesh in the first object parallel thread resource or the parallel thread resource
Parallel thread resource is marked, to execute the corresponding parallel computation process of identical fine granularity.
Second aspect, the embodiment of the present invention provide a kind of device that extensive continuous tide calculates, and described device includes：
First parallel computation unit, for being each operating condition point in the work information according to the work information being obtained ahead of time
With corresponding parallel thread resource, with the corresponding load level of each operating condition described in parallel computation；
Second parallel computation unit, for being institute in the parallel thread resource according to predetermined stressor
The corresponding default load level distribution first object parallel thread resource of each of each operating condition operating condition is stated, with parallel computation institute
State default load level；
Third parallel computation unit, in the process for calculating the default load level or the load level every time
In, obtain the fine granularity of calculating, be identical fine-grained calculating process the first object parallel thread resource or it is described simultaneously
The second target parallel thread resources are distributed in row thread resources, to execute the corresponding parallel computation process of identical fine granularity.
The third aspect, the embodiment of the present invention provide a kind of electronic equipment, including：Processor, memory and bus, wherein
The processor and the memory complete mutual communication by the bus；
The memory is stored with the program instruction that can be executed by the processor, and the processor calls described program to refer to
Order is able to carry out following method：
Corresponding parallel thread resource is distributed for each operating condition in the work information according to the work information being obtained ahead of time,
With the corresponding load level of each operating condition described in parallel computation；
It is each of each operating condition operating condition in the parallel thread resource according to predetermined stressor
Corresponding default load level distributes first object parallel thread resource, to preset load level described in parallel computation；
During calculating the default load level or the load level every time, the fine granularity of calculating is obtained, is
Identical fine-grained calculating process distributes the second mesh in the first object parallel thread resource or the parallel thread resource
Parallel thread resource is marked, to execute the corresponding parallel computation process of identical fine granularity.
Fourth aspect, the embodiment of the present invention provide a kind of non-transient computer readable storage medium, including：
The non-transient computer readable storage medium stores computer instruction, and the computer instruction makes the computer
Execute following method：
Corresponding parallel thread resource is distributed for each operating condition in the work information according to the work information being obtained ahead of time,
With the corresponding load level of each operating condition described in parallel computation；
It is each of each operating condition operating condition in the parallel thread resource according to predetermined stressor
Corresponding default load level distributes first object parallel thread resource, to preset load level described in parallel computation；
During calculating the default load level or the load level every time, the fine granularity of calculating is obtained, is
Identical fine-grained calculating process distributes the second mesh in the first object parallel thread resource or the parallel thread resource
Parallel thread resource is marked, to execute the corresponding parallel computation process of identical fine granularity.
The method and device that extensive continuous tide provided in an embodiment of the present invention calculates, by distributing at many levels parallel
Thread resources can be improved the computational efficiency of continuous tide calculating, to mention to carry out the parallel computation of extensive continuous tide
Rise the efficiency and accuracy of large scale system Voltage stability analysis.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is this hair
Bright some embodiments for those of ordinary skill in the art without creative efforts, can be with root
Other attached drawings are obtained according to these attached drawings.
Fig. 1 is a kind of method flow schematic diagram that extensive continuous tide calculates of the embodiment of the present invention；
Fig. 2 is the parallel algorithm schematic diagram for the multi-layer that the extensive continuous tide of the embodiment of the present invention calculates；
Fig. 3 is prior art continuous tide calculation method schematic diagram；
Fig. 4 is continuous tide of embodiment of the present invention calculation method schematic diagram；
Fig. 5 is level of the embodiment of the present invention 1 and more Load flow calculation scene figures that 2 parallel computation of level provides；
Fig. 6 is pushed forward back substitution based on the batch of LDAG for the embodiment of the present invention and calculates schematic diagram；
Fig. 7 is the apparatus structure schematic diagram that the extensive continuous tide of the embodiment of the present invention calculates；
Fig. 8 is electronic equipment entity structure schematic diagram provided in an embodiment of the present invention.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is
A part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art
Every other embodiment obtained without creative efforts, shall fall within the protection scope of the present invention.
Embodiment to facilitate the understanding of the present invention is first explained as follows relational language：
1. continuous tide：Continuous tide (Continuation Power Flow, CPF) is also known as continuation power flow, is electric power
The powerful of system voltage stability analysis.
2.GPU：Full name Graphics Processing Unit, graphics processor have massive concurrent thread, as simultaneously
Row calculates carrier and is widely used in the acceleration calculating of every field.
3.CUDA：Full name Compute Unified Device Architecture is that one kind is released by NVIDIA company
Universal parallel computing architecture, provide the multi-level programming model that can be developed using C language.
Fig. 1 is a kind of method flow schematic diagram that extensive continuous tide calculates of the embodiment of the present invention, as shown in Figure 1, this
The method that a kind of extensive continuous tide that inventive embodiments provide calculates, includes the following steps：
S101：According to the corresponding parallel thread of each operating condition distribution that the work information being obtained ahead of time is in the work information
Resource, with the corresponding load level of each operating condition described in parallel computation.
Specifically, device is corresponding simultaneously for each operating condition distribution in the work information according to the work information being obtained ahead of time
Row thread resources, with the corresponding load level of each operating condition described in parallel computation.Device can be server, terminal etc. and carry
The carrier of GPU, this method execute in GPU.Fig. 2 is the parallel of the multi-layer that the extensive continuous tide of the embodiment of the present invention calculates
Algorithm schematic diagram, multi-layer as shown in Figure 2 are divided into that multi-state is parallel, more trend sections are parallel, single trend fine grained parallel
Three levels.
Level 1 is described as follows：
It is i.e. parallel in the case of consideration generator and load fluctuation.Scene is grid-connected to have generator output centainly not
Certainty, the usually random fluctuation in a certain range of the active power output for the node that generates electricity；And the access of the novel loads such as electric car
The disturbance situation that may also appear in electric load in the short time.These all make electric system actual motion state compared with
It will appear various working in short time, need to be compared various working comprehensive continuous tide and calculate, it could be to current fortune
Voltage stability under row state has comprehensive and accurate judgement, to avoid the generation of collapse of voltage accident.The company of different operating conditions
Continuous Load flow calculation is brought at double again in calculation amount before, is difficult to receive for large scale system.It can be found through analysis
In the immovable situation of system topological, the continuous tide calculating between each operating condition has no any data dependence, therefore can design simultaneously
Row algorithm carries out acceleration calculating.
Illustrate the preparatory acquisition methods of work information below：
The fluctuation that the embodiment of the present invention may be occurred using the method simulation generator and load that are randomly provided the fluctuation factor.
Assuming that have r generator node, s load bus in node system, then generator node active power output, load bus it is active
It can be arranged as (r+2*s) dimensional vector k according to node serial number with idle, be called generation load vector.
The fluctuation of construction (r+2*s) dimension is because of subvector α according to this, wherein preceding r element represents the wave of r power plant active power output
Reason, rear 2*s element represent the active and reactive power fluctuation factor of s load.For basic example, any wave does not occur
It is dynamic, then there are α i=1, i=1,2 ..., r+2*s, k=k(0)For basic generation load vector.It is certain to correspond to new energy based on this
Source power plant with can the fluctuation factor-alpha i value of fluctuating load can change in a certain range, generate different fluctuations because of subvector,
Correspond respectively to different operating conditions.Assuming that considering m kind operating condition, then m different vector α constitute (r+2*s) * m dimension fluctuation
Factor matrix A={ aij, wherein aijIndicate i-th of (i under j-th of operating condition<=r) generator node or (i-r) a (r<i<
=r+s) load bus the fluctuation factor, be [amin,amax] between random number, aminAnd amaxFor preset constant, fluctuation is represented
Factor bound.
Different work informations can be obtained by formula (1), wherein k(0) iFor i-th of element of basic generation load vector, k(j) iFor i-th of element of acquired jth operating condition generation load vector.
k(j) i=aij·k(0) i (1)
It can be expressed as shown in formula (2) with matrix form, wherein K is to finally obtain (the r+2* comprising all operating conditions
S) * m ties up initial generation load matrix, wherein jth column vector is the initial generation load vector of operating condition j；K(0)=diag (k(0)) it is the diagonal matrix formed by basic generation load vector.
K=K(0)*A (2)
The each column of matrix K correspond to each work information, as initial generated power and are born using every column data
Lotus active and reactive power value, the then continuous tide that can be carried out under m kind operating condition calculate.Under CUDA framework, matrix K can be by electric power
System data precalculates to obtain, and each column data is distributed to certain parallel thread resource later, may include thread block
And thread, the continuous tide calculating between different lines can start simultaneously at, the as parallel computation between multi-state.
S102：It is each in each operating condition in the parallel thread resource according to predetermined stressor
The corresponding default load level of a operating condition distributes first object parallel thread resource, to preset load level described in parallel computation.
Specifically, device is in each operating condition in the parallel thread resource according to predetermined stressor
The corresponding default load level of each operating condition distribute first object parallel thread resource, to preset load described in parallel computation
It is horizontal.
Level 2 is described as follows referring to Fig. 2：
Level 2 is parallel, i.e., the Load flow calculation under the corresponding default load level of each operating condition of more trend sections
Parallel.The level changes traditional continuous tide and calculates mode from low load levels to high load capacity level successively Continuous plus,
Directly the trend section under different load level is solved, judges the attainable limit point of load level with convergence,
Embody the basic gender gap of algorithm of the embodiment of the present invention Yu existing continuous power flow.Fig. 3 is prior art continuous tide meter
Method schematic diagram is calculated, as shown in figure 3, the method for the prior art is successively serial to ill horizontal E point for the A point of basal level
It calculates, the abscissa (i.e. load level) of each point is calculated by a upper point.Fig. 4 is continuous tide of embodiment of the present invention meter
Method schematic diagram is calculated, as shown in figure 4, the method for the embodiment of the present invention is directly according to a series of given load levels to A-F point
Carry out Load flow calculation simultaneously, A-E point can restrain and F point cannot, then will convergent maximum load level as load pole
Point of accumulation.
It should be noted that：Fig. 4 illustrates to use the PV curve (power vs. voltage curve) as Fig. 3 only for facilitating
With mark point (i.e. A~F in figure).In fact, parallel method is since abscissa is in advance to give, it can't be with the side of the prior art
Method is completely coincident on calculating data point.The mode that mean allocation generally can be used gives one group of stressor sequence λ1~λL,
In, x-th of stressor λxFor：
Wherein, λxFor x-th of stressor, 0 < x≤L；λlowIt is general desirable for preset calculating starting point stressor
1.0；λupIt is horizontal for preset peak load, it can generally be estimated as 3.0,4.0,5.0 equivalences, must be greater than according to real system situation
Load limit point (the load limit point determined referring to Fig. 4) at system practical limit point；L is default stressor number
(exemplary A~E is 5 total in corresponding diagram 2), the more big then data point distribution of value is more intensive, and accuracy is higher, but calculation amount
Corresponding increase can select suitable numerical value according to hardware resource and required precision.
Different trend sections corresponds to different load levels, and each load level exactly above-mentioned generation load to
It is obtained on the basis of amount multiplied by stressor.Therefore, in given one group of stressor sequence λ1~λLIn the case where, according to first
Electric load matrix K is originated, it is in need applied in Load flow calculation that the institute under m kind operating condition and L default load levels can be obtained
Generation load vector, amount to m*L, wherein the generation load vector k of (i*m+x) a default load level* (i*m+x)By formula
(4) it is calculated, wherein k* (i*m+x)For generation load vector, the λ of (i*m+x) a default load levelxFor x-th of load
The factor, kiFor i-th in (r+2*s) dimensional vector k.
k* (i*m+x)=λxki (4)
Fig. 5 is level of the embodiment of the present invention 1 and more Load flow calculation scene figures that 2 parallel computation of level provides, such as Fig. 5 institute
To show, the design of preceding two layers of parallel calculating method is actually to provide a series of Load flow calculation scene of non-correlations, these
System network architecture is all the same in scape, but generated power power output is different with load, therefore calculation of tidal current is not also identical.Needle
The characteristics of possessing massive concurrent thread to GPU, reasonable distribution computing resource carry out batch Parallel implementation to these trend scenes, then
PV curve and load limit point, load margin under different operating conditions can be obtained, to complete the Voltage stability analysis of system.
The parallel calculating method of more trend sections is to adapt to parallel architecture and abandon in existing method some important patrols
Volume, such as contacting between iterative initial value and the calculating of a upper load level and the parameterized treatment to avoid Singular Value from using
Deng, single tidal current computing method certainly will be required to be capable of handling these special circumstances, therefore, the embodiment of the present invention use continuous ox
The tidal current computing method for the method for pausing.This method can be suitable for simultaneously ground state in zero start and Abnormal Load Flow of Power Systems solves, this guarantor
The successful realization of 2 parallel computation of level is hindered.
S103：During calculating the default load level or the load level every time, the particulate of calculating is obtained
Degree distributes for identical fine-grained calculating process in the first object parallel thread resource or the parallel thread resource
Two target parallel thread resources, to execute the corresponding parallel computation process of identical fine granularity.
Specifically, device during calculating the default load level or the load level every time, is obtained and is calculated
Fine granularity, be identical fine-grained calculating process in the first object parallel thread resource or the parallel thread resource
The second target parallel thread resources are distributed, to execute the corresponding parallel computation process of identical fine granularity.
It is described as follows that (the case where during calculating the load level every time is not shown, together to level 3 referring to Fig. 2
The case where managing, being referred to during calculating the default load level every time)：
Level 3 is the fine grained parallel of single Load flow calculation.Preceding two layers of parallel algorithm is by the continuous tide meter under multi-state
The common Load flow calculation for being split as multiple no dependences is calculated, is a kind of coarse grain parallelism of task rank, although comparing serial algorithm
There is higher efficiency, but be unable to fully computation resource, improved efficiency is restricted.In conjunction with the magnanimity on high-performance calculation GPU
Thread computing resource should be directed to single Load flow calculation Process Design fine granularity parallel algorithm, and it is right inside Kernel to calculate in GPU
Thread structure is reasonably adjusted, and call duration time is reduced, and realizes that efficiency maximizes.
The embodiment of the present invention solves using the continuous Newton Power Flow method for solving for determining Jacobi, reply power flow equation
It wherein mainly calculates step to be analyzed, designs corresponding batch parallel algorithm.Its mainly calculate include Jacobian matrix LU decompose,
Trend imbalance power amount f (x) is solved, is pushed forward back substitution calculating and the update of state variable.Wherein LU is decomposed to fixed example
It need to only carry out once, this makes LU decomposition that can be previously-completed and form factor table for applying in iteration, and calculation amount is in entirety
Account for smaller in calculating, simultaneously because LU decomposable process includes more complicated logic, therefore it is parallel for LU not decompose design
Algorithm, it is serial in advance in calculating to complete.The calculating of other three parts need to carry out repeatedly in iteration, and according to different calculating scenes
And change, design fine granularity parallel algorithm is completed.
Trend imbalance power is solved, actually a series of voltage amount, admittance matrix real part imaginary part and phase angle
The product cycle of sine and cosine value is summed.It, can be according to the non-zero in admittance matrix in conjunction with the sparsity of electric system admittance matrix
Summation process is split as the multiple additive process of no data dependence by element, and configuration relevant calculation Resource Design fine grained parallel is calculated
Method.It should be noted that this algorithm is related to the operation that different computing resources are summed up and are written into unified memory address,
It will appear warfare, therefore can use atom add operation and realize circulation summation.
For be pushed forward back substitution calculate, actually coefficient matrix be upper triangular matrix and lower triangular matrix sparse linear side
Journey group solution procedure can uniformly be attributed to the update and normalization of node corresponding element value.The process has apparent serial suitable
Sequence estimated performance, the node elements value updated afterwards depend on the node for first completing to update, it is impossible to realize complete parallel.But it is this
Sequential dependency be still it is sparse, i.e., every time calculate be solely dependent upon before calculate once or several times as a result, this is allowed for wherein
A large amount of non-correlation calculating still are able to complete parallel.In view of this, calculation process is split as point being made of basic calculating member
Layer digraph completes fine granularity calculating with interlayer parallel.
In addition, the more new content of state variable is the add operation of wherein each element.More due to different conditions variable
Between new and onrelevant, the Parallel Implementation that state variable updates is relatively simple, distributes corresponding computing resource to every in state variable
A element, while parallel completion.
Fig. 6 is pushed forward back substitution based on the batch of LDAG for the embodiment of the present invention and calculates schematic diagram, and LDAG therein is oriented nothing
Ring figure layer (Layered Directed Acyclic Graph), as shown in fig. 6, being directed to the feelings that hierarchical directed graph is m
Condition repeats no more in the case of hierarchical directed graph is L.1~scene of scene m in Fig. 6 respectively corresponds above-mentioned 1~work of operating condition
The corresponding serial thread of condition m, Thread1~Threadn, such as：The corresponding basic calculating member 2 of Thread1, basic calculating member 3,
It is serial computing thread between basic calculating member 8, basic calculating member 10 and basic calculating member 11, Block1~Blockm is corresponding
Be parallel thread resource, Lever1~Lever h it is corresponding be hierarchical directed graph each layer, by taking Block1 as an example,
H the second target parallel thread resources are distributed in the corresponding parallel thread resource of Block1, i.e. the in Block1~Blockm
One layer (total m) occupies second target parallel thread resources Y1, simultaneously and concurrently calculate basic calculating member 2, basic calculating member 5,
Basic calculating member 1, basic calculating member 4；The second layer (total m) in Block1~Blockm occupies the second target parallel thread
Resource Y2 simultaneously and concurrently calculates basic calculating member 3, basic calculating member 6, and so on.Referring to the example above, identical fine granularity is
Identical layer in corresponding m scene.
The method that extensive continuous tide provided in an embodiment of the present invention calculates is provided by distributing multi-level parallel thread
Source can be improved the computational efficiency of continuous tide calculating to carry out the parallel computation of extensive continuous tide, to promote big rule
The efficiency and accuracy of modular system Voltage stability analysis.
On the basis of the above embodiments, the work information is obtained ahead of time, including：
Obtain generation load vector；The generation load vector includes a negative by the active power output of r generator node, s
(r+2*s) the dimensional vector k of the active and idle composition of lotus node.
Specifically, device obtains generation load vector；The generation load vector includes by the active of r generator node
Power output, s load bus active and idle composition (r+2*s) dimensional vector k.It can refer to above-described embodiment, repeat no more.
According to preset m kind operating condition, the fluctuation factor matrix A of (r+2*s) * m dimension is constructed；Wherein, (r+2*s) is indicated
The line number of A, m indicate that the columns of A, the element in A are aij；aijIt is [amin,amax] between random number, aminAnd amaxIt is default
Constant.
Specifically, device constructs the fluctuation factor matrix A of (r+2*s) * m dimension according to preset m kind operating condition；Wherein,
(r+2*s) indicate that the line number of A, m indicate that the columns of A, the element in A are aij；aijIt is [amin,amax] between random number, amin
And amaxFor preset constant.It can refer to above-described embodiment, repeat no more.
According to the k and the A, initial generation load matrix K is obtained, wherein each column of the K correspond to each work
Condition information.
Specifically, device obtains initial generation load matrix K according to the k and the A, wherein each column of the K
Each corresponding work information.It can refer to above-described embodiment, repeat no more.
The method that extensive continuous tide provided in an embodiment of the present invention calculates can rationally and effectively get operating condition letter
Breath.
On the basis of the above embodiments, the work information that the basis is obtained ahead of time is each work in the work information
Condition distributes corresponding parallel thread resource, including：
According to the operating condition species number m of the work information, distributes m and provided with each one-to-one parallel thread of operating condition
Source.
Specifically, operating condition species number m of the device according to the work information, distribution m correspond with each operating condition
Parallel thread resource.It can refer to above-described embodiment, repeat no more.
The method that extensive continuous tide provided in an embodiment of the present invention calculates, it is one-to-one with each operating condition by distributing
Parallel thread resource ensure that being normally carried out for multi-state level parallel computation, be further able to improve what continuous tide calculated
Computational efficiency, to promote the efficiency and accuracy of large scale system Voltage stability analysis.
On the basis of the above embodiments, the stressor is predetermined, including：
The stressor is calculated according to the following formula：
Wherein, λxFor x-th of stressor, 0 < x≤L；λlowFor preset calculating starting point stressor；λupIt is default
Peak load it is horizontal；L is default stressor number.
Specifically, device calculates the stressor according to the following formula：
Wherein, λxFor x-th of stressor, 0 < x≤L；λlowFor preset calculating starting point stressor；λupIt is default
Peak load it is horizontal；L is default stressor number.It can refer to above-described embodiment, repeat no more.
The method that extensive continuous tide provided in an embodiment of the present invention calculates, can rationally and effectively determine load because
Son.
On the basis of the above embodiments, described according to predetermined stressor, in the parallel thread resource
First object parallel thread resource is distributed for the corresponding default load level of each of each operating condition operating condition, including：
According to the L, L and the default load level one-to-one first are distributed in the parallel thread resource
Target parallel thread resources.
Specifically, device according to the L, distributes L and the default load level one in the parallel thread resource
One corresponding first object parallel thread resource.It can refer to above-described embodiment, repeat no more.
The method that extensive continuous tide provided in an embodiment of the present invention calculates, by distributing with default load level one by one
Corresponding first object parallel thread resource, ensure that being normally carried out for more trend section parallel computations, be further able to improve
The computational efficiency that continuous tide calculates, to promote the efficiency and accuracy of large scale system Voltage stability analysis.
On the basis of the above embodiments, described to preset load level described in parallel computation, including：
According to the following formula to preset load level described in parallel computation：
k* (i*m+x)=λxki
Wherein, k* (i*m+x)For generation load vector, the λ of (i*m+x) a default load levelxFor x-th of stressor,
kiFor i-th in (r+2*s) dimensional vector k.
Specifically, device is according to the following formula to preset load level described in parallel computation：
k* (i*m+x)=λxki
Wherein, k* (i*m+x)For generation load vector, the λ of (i*m+x) a default load levelxFor x-th of stressor,
kiFor i-th in (r+2*s) dimensional vector k.It can refer to above-described embodiment, repeat no more.
The method that extensive continuous tide provided in an embodiment of the present invention calculates can be accurate, fast by specific formula
Default load level is calculated fastly.
On the basis of the above embodiments, the default load level or institute are calculated using being pushed forward back substitution tidal current computing method
State load level；Correspondingly, described is identical fine-grained calculating process in the first object parallel thread resource or described
The second target parallel thread resources are distributed in parallel thread resource, including：
The process for being pushed forward back substitution Load flow calculation is split as include basic calculating member hierarchical directed graph；Wherein, described
Hierarchical directed graph is m or L.
Specifically, device by the process for being pushed forward back substitution Load flow calculation be split as include basic calculating member layering it is oriented
Figure；Wherein, the hierarchical directed graph is m or L.It can refer to above-described embodiment, repeat no more.
If the hierarchical directed graph is m, h the second target parallel thread moneys are distributed in the parallel thread resource
Source, so that each identical layer in m hierarchical directed graph occupies corresponding each second target parallel thread resources respectively；
Wherein, the h is the number of plies of the hierarchical directed graph.
If distributing h in the parallel thread resource specifically, device judges to know that the hierarchical directed graph is m
A second target parallel thread resources so that each identical layer in m hierarchical directed graph occupy respectively it is corresponding each described
Second target parallel thread resources；Wherein, the h is the number of plies of the hierarchical directed graph.It can refer to above-described embodiment, it is no longer superfluous
It states.
Or,
If the hierarchical directed graph is L, h the second targets are distributed in the first object parallel thread resource simultaneously
Row thread resources, so that each identical layer in L hierarchical directed graph occupies corresponding each second target parallel respectively
Thread resources；Wherein, the h is the number of plies of the hierarchical directed graph.
Specifically, if device judges to know that the hierarchical directed graph is L, in the first object parallel thread resource
Middle distribution h the second target parallel thread resources so that each identical layer in L hierarchical directed graph occupy respectively it is corresponding often
The one second target parallel thread resources；Wherein, the h is the number of plies of the hierarchical directed graph.It can refer to above-described embodiment,
It repeats no more.
The method that extensive continuous tide provided in an embodiment of the present invention calculates, passes through each phase into hierarchical directed graph
With Layer assignment the second target parallel thread resources, single trend fine granularity is determined as the identical layer in hierarchical directed graph, is gone forward side by side
Row parallel computation is further able to improve the computational efficiency that continuous tide calculates, to promote large scale system voltage stability
The efficiency and accuracy of analysis.
Fig. 7 is the apparatus structure schematic diagram that the extensive continuous tide of the embodiment of the present invention calculates, as shown in fig. 7, of the invention
Embodiment provides a kind of device that extensive continuous tide calculates, including the first parallel computation unit 701, the first parallel computation
Unit 702 and the first parallel computation unit 703, wherein：
First parallel computation unit 701 is used to be each operating condition in the work information according to the work information being obtained ahead of time
Corresponding parallel thread resource is distributed, with the corresponding load level of each operating condition described in parallel computation；Second parallel computation unit 702
For being that each of each operating condition operating condition is corresponding in the parallel thread resource according to predetermined stressor
Default load level distribute first object parallel thread resource, with described in parallel computation preset load level；Third is counted parallel
It calculates unit 703 to be used for during calculating the default load level or the load level every time, obtains the particulate of calculating
Degree distributes for identical fine-grained calculating process in the first object parallel thread resource or the parallel thread resource
Two target parallel thread resources, to execute the corresponding parallel computation process of identical fine granularity.
Specifically, the first parallel computation unit 701 is used to according to the work information being obtained ahead of time be in the work information
Each operating condition distribute corresponding parallel thread resource, with the corresponding load level of each operating condition described in parallel computation；Second parallel meter
It calculates unit 702 to be used for according to predetermined stressor, is each in each operating condition in the parallel thread resource
The corresponding default load level of a operating condition distributes first object parallel thread resource, to preset load level described in parallel computation；
Third parallel computation unit 703 is used for during calculating the default load level or the load level every time, is obtained
The fine granularity of calculating is that identical fine-grained calculating process is provided in the first object parallel thread resource or the parallel thread
The second target parallel thread resources are distributed in source, to execute the corresponding parallel computation process of identical fine granularity.
The device that extensive continuous tide provided in an embodiment of the present invention calculates is provided by distributing multi-level parallel thread
Source can be improved the computational efficiency of continuous tide calculating to carry out the parallel computation of extensive continuous tide, to promote big rule
The efficiency and accuracy of modular system Voltage stability analysis.
The device that extensive continuous tide provided in an embodiment of the present invention calculates specifically can be used for executing above-mentioned each method
The process flow of embodiment, details are not described herein for function, is referred to the detailed description of above method embodiment.
Fig. 8 is electronic equipment entity structure schematic diagram provided in an embodiment of the present invention, as shown in figure 8, the electronic equipment
Including：Processor (processor) 801, memory (memory) 802 and bus 803；
Wherein, the processor 801, memory 802 complete mutual communication by bus 803；
The processor 801 is used to call the program instruction in the memory 802, to execute above-mentioned each method embodiment
Provided method, for example including：According to each operating condition distribution correspondence that the work information being obtained ahead of time is in the work information
Parallel thread resource, with the corresponding load level of each operating condition described in parallel computation；According to predetermined stressor, in institute
State in parallel thread resource is that the corresponding default load level distribution first object of each of each operating condition operating condition is parallel
Thread resources, to preset load level described in parallel computation；The default load level or the load level are being calculated every time
During, obtain the fine granularity of calculating, be identical fine-grained calculating process in the first object parallel thread resource or
The second target parallel thread resources are distributed in the parallel thread resource, to execute the corresponding parallel computation of identical fine granularity
Journey.
The present embodiment discloses a kind of computer program product, and the computer program product includes being stored in non-transient calculating
Computer program on machine readable storage medium storing program for executing, the computer program include program instruction, when described program instruction is calculated
When machine executes, computer is able to carry out method provided by above-mentioned each method embodiment, for example including：According to the work being obtained ahead of time
Condition information is that each operating condition in the work information distributes corresponding parallel thread resource, corresponding with each operating condition described in parallel computation
Load level；It is each of described each operating condition in the parallel thread resource according to predetermined stressor
The corresponding default load level of operating condition distributes first object parallel thread resource, to preset load level described in parallel computation；?
During calculating the default load level or the load level every time, the fine granularity of calculating is obtained, is identical fine granularity
Calculating process distribute the second target parallel thread in the first object parallel thread resource or the parallel thread resource
Resource, to execute the corresponding parallel computation process of identical fine granularity.
The present embodiment provides a kind of non-transient computer readable storage medium, the non-transient computer readable storage medium
Computer instruction is stored, the computer instruction makes the computer execute method provided by above-mentioned each method embodiment, example
Such as include：Corresponding parallel thread resource is distributed for each operating condition in the work information according to the work information being obtained ahead of time,
With the corresponding load level of each operating condition described in parallel computation；According to predetermined stressor, in the parallel thread resource
In be the corresponding default load level distribution first object parallel thread resource of each of each operating condition operating condition, with parallel
Calculate the default load level；During calculating the default load level or the load level every time, meter is obtained
The fine granularity of calculation is identical fine-grained calculating process in the first object parallel thread resource or the parallel thread resource
The second target parallel thread resources of middle distribution, to execute the corresponding parallel computation process of identical fine granularity.
Those of ordinary skill in the art will appreciate that：Realize that all or part of the steps of above method embodiment can pass through
The relevant hardware of program instruction is completed, and program above-mentioned can be stored in a computer readable storage medium, the program
When being executed, step including the steps of the foregoing method embodiments is executed；And storage medium above-mentioned includes：ROM, RAM, magnetic disk or light
The various media that can store program code such as disk.
The embodiments such as electronic equipment described above are only schematical, wherein it is described as illustrated by the separation member
Unit may or may not be physically separated, and component shown as a unit may or may not be object
Manage unit, it can it is in one place, or may be distributed over multiple network units.It can select according to the actual needs
Some or all of the modules therein is selected to achieve the purpose of the solution of this embodiment.Those of ordinary skill in the art are not paying wound
In the case where the labour for the property made, it can understand and implement.
Through the above description of the embodiments, those skilled in the art can be understood that each embodiment can
It realizes by means of software and necessary general hardware platform, naturally it is also possible to pass through hardware.Based on this understanding, on
Stating technical solution, substantially the part that contributes to existing technology can be embodied in the form of software products in other words, should
Computer software product may be stored in a computer readable storage medium, such as ROM/RAM, magnetic disk, CD, including several fingers
It enables and using so that a computer equipment (can be personal computer, server or the network equipment etc.) executes each implementation
Method described in certain parts of example or embodiment.
Finally it should be noted that：The above various embodiments is only to illustrate the technical solution of the embodiment of the present invention, rather than right
It is limited；Although the embodiment of the present invention is described in detail referring to foregoing embodiments, the ordinary skill of this field
Personnel should understand that：It is still possible to modify the technical solutions described in the foregoing embodiments, or to part
Or all technical features are equivalently replaced；And these are modified or replaceed, it does not separate the essence of the corresponding technical solution
The range of various embodiments of the present invention technical solution.
Claims (10)
1. a kind of method that extensive continuous tide calculates, which is characterized in that including：
According to the corresponding parallel thread resource of each operating condition distribution that the work information being obtained ahead of time is in the work information, with simultaneously
Row calculates the corresponding load level of each operating condition；
It is that each of each operating condition operating condition is corresponding in the parallel thread resource according to predetermined stressor
Default load level distribute first object parallel thread resource, with described in parallel computation preset load level；
During calculating the default load level or the load level every time, the fine granularity of calculating is obtained, is identical
Fine-grained calculating process distributes the second target simultaneously in the first object parallel thread resource or the parallel thread resource
Row thread resources, to execute the corresponding parallel computation process of identical fine granularity.
2. the method according to claim 1, wherein the work information is obtained ahead of time, including：
Obtain generation load vector；The generation load vector includes active power output by r generator node, s load section
(r+2*s) the dimensional vector k of the active and idle composition of point；
According to preset m kind operating condition, the fluctuation factor matrix A of (r+2*s) * m dimension is constructed；Wherein, (r+2*s) indicates A's
Line number, m indicate that the columns of A, the element in A are aij；aijIt is [amin,amax] between random number, aminAnd amaxIt is default normal
Number；
According to the k and the A, initial generation load matrix K is obtained, wherein each column of the K correspond to each operating condition letter
Breath.
3. according to the method described in claim 2, it is characterized in that, the work information that the basis is obtained ahead of time is the operating condition
Each operating condition in information distributes corresponding parallel thread resource, including：
According to the operating condition species number m of the work information, m and each one-to-one parallel thread resource of operating condition are distributed.
4. according to the method described in claim 3, it is characterized in that, the stressor predefine, including：
The stressor is calculated according to the following formula：
Wherein, λxFor x-th of stressor, 0 < x≤L；λlowFor preset calculating starting point stressor；λupFor it is preset most
Big load level；L is default stressor number.
5. according to the method described in claim 4, it is characterized in that, described according to predetermined stressor, it is described simultaneously
It is the corresponding default load level distribution first object parallel thread of each of each operating condition operating condition in row thread resources
Resource, including：
According to the L, L and the default one-to-one first object of load level are distributed in the parallel thread resource
Parallel thread resource.
6. according to the method described in claim 4, it is characterized in that, it is described with described in parallel computation preset load level, including：
According to the following formula to preset load level described in parallel computation：
k* (i*m+x)=λxki
Wherein, k* (i*m+x)For generation load vector, the λ of (i*m+x) a default load levelxFor x-th of stressor, kiFor
(r+2*s) i-th in dimensional vector k.
7. according to the method described in claim 4, it is characterized in that, calculating described preset using back substitution tidal current computing method is pushed forward
Load level or the load level；Correspondingly, described is identical fine-grained calculating process in the first object parallel line
The second target parallel thread resources are distributed in Cheng Ziyuan or the parallel thread resource, including：
The process for being pushed forward back substitution Load flow calculation is split as include basic calculating member hierarchical directed graph；Wherein, the layering
Digraph is m or L；
If the hierarchical directed graph is m, h the second target parallel thread resources are distributed in the parallel thread resource,
So that each identical layer in m hierarchical directed graph occupies corresponding each second target parallel thread resources respectively；Its
In, the h is the number of plies of the hierarchical directed graph；
Or,
If the hierarchical directed graph is L, h the second target parallel lines are distributed in the first object parallel thread resource
Cheng Ziyuan, so that each identical layer in L hierarchical directed graph occupies corresponding each second target parallel thread respectively
Resource；Wherein, the h is the number of plies of the hierarchical directed graph.
8. a kind of device that extensive continuous tide calculates, which is characterized in that including：
First parallel computation unit, for being each operating condition distribution pair in the work information according to the work information being obtained ahead of time
The parallel thread resource answered, with the corresponding load level of each operating condition described in parallel computation；
Second parallel computation unit, for being described each in the parallel thread resource according to predetermined stressor
The corresponding default load level of each of operating condition operating condition distributes first object parallel thread resource, with pre- described in parallel computation
If load level；
Third parallel computation unit, for obtaining during calculating the default load level or the load level every time
The fine granularity of calculating is taken, is identical fine-grained calculating process in the first object parallel thread resource or the parallel thread
The second target parallel thread resources are distributed in resource, to execute the corresponding parallel computation process of identical fine granularity.
9. a kind of electronic equipment, which is characterized in that including：Processor, memory and bus, wherein
The processor and the memory complete mutual communication by the bus；
The memory is stored with the program instruction that can be executed by the processor, and the processor calls described program to instruct energy
Enough methods executed as described in claim 1 to 7 is any.
10. a kind of non-transient computer readable storage medium, which is characterized in that the non-transient computer readable storage medium is deposited
Computer instruction is stored up, the computer instruction makes the computer execute the method as described in claim 1 to 7 is any.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810646054.3A CN108879691B (en) | 2018-06-21 | 2018-06-21 | Large-scale continuous power flow calculation method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810646054.3A CN108879691B (en) | 2018-06-21 | 2018-06-21 | Large-scale continuous power flow calculation method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108879691A true CN108879691A (en) | 2018-11-23 |
| CN108879691B CN108879691B (en) | 2020-09-04 |
Family
ID=64340173
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810646054.3A Active CN108879691B (en) | 2018-06-21 | 2018-06-21 | Large-scale continuous power flow calculation method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108879691B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110704023A (en) * | 2019-09-26 | 2020-01-17 | 北京华大九天软件有限公司 | Matrix block division method and device based on topological sorting |
| CN113672775A (en) * | 2021-08-06 | 2021-11-19 | 南方电网数字电网研究院有限公司 | Parallel computing method for real-time graph of power grid |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102593820A (en) * | 2011-12-22 | 2012-07-18 | 河海大学 | Continuation power flow algorithm considering field current constraint and armature current constraint of power generator |
| CN104156609A (en) * | 2014-08-19 | 2014-11-19 | 河海大学 | Power grid flow solvability identification and initial value generating method based on continuous power flow |
| CN105391057A (en) * | 2015-11-20 | 2016-03-09 | 国家电网公司 | GPU thread design method of power flow Jacobian matrix calculation |
| CN105576648A (en) * | 2015-11-23 | 2016-05-11 | 中国电力科学研究院 | Static security analysis double-layer parallel method based on GPU-CUP heterogeneous computing platform |
| CN106874113A (en) * | 2017-01-19 | 2017-06-20 | 国电南瑞科技股份有限公司 | A kind of many GPU heterogeneous schemas static security analysis computational methods of CPU+ |
-
2018
- 2018-06-21 CN CN201810646054.3A patent/CN108879691B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102593820A (en) * | 2011-12-22 | 2012-07-18 | 河海大学 | Continuation power flow algorithm considering field current constraint and armature current constraint of power generator |
| CN104156609A (en) * | 2014-08-19 | 2014-11-19 | 河海大学 | Power grid flow solvability identification and initial value generating method based on continuous power flow |
| CN105391057A (en) * | 2015-11-20 | 2016-03-09 | 国家电网公司 | GPU thread design method of power flow Jacobian matrix calculation |
| CN105576648A (en) * | 2015-11-23 | 2016-05-11 | 中国电力科学研究院 | Static security analysis double-layer parallel method based on GPU-CUP heterogeneous computing platform |
| CN106874113A (en) * | 2017-01-19 | 2017-06-20 | 国电南瑞科技股份有限公司 | A kind of many GPU heterogeneous schemas static security analysis computational methods of CPU+ |
Non-Patent Citations (3)
| Title |
|---|
| VINCENT ROBERGE等: "Parallel Power Flow on Graphics Processing Units for Concurrent Evaluation of Many Networks", 《IEEE TRANSACTIONS ON SMART GRID》 * |
| 王宗杰等: "面向时间过程的特征最优潮流模型", 《电力系统自动化》 * |
| 王明轩等: "适用于ＣＰＵ＋ＧＰＵ协同架构的大规模病态潮流求解方法", 《电力系统自动化》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110704023A (en) * | 2019-09-26 | 2020-01-17 | 北京华大九天软件有限公司 | Matrix block division method and device based on topological sorting |
| CN110704023B (en) * | 2019-09-26 | 2021-10-22 | 北京华大九天科技股份有限公司 | Matrix block division method and device based on topological sorting |
| CN113672775A (en) * | 2021-08-06 | 2021-11-19 | 南方电网数字电网研究院有限公司 | Parallel computing method for real-time graph of power grid |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108879691B (en) | 2020-09-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| He et al. | Probability density forecasting of wind power based on multi-core parallel quantile regression neural network | |
| CN104836256B (en) | The method and system that a kind of power distribution network photovoltaic digestion capability calculates | |
| Hardiansyah et al. | Solving economic load dispatch problem using particle swarm optimization technique | |
| Shilaja et al. | Internet of medical things-load optimization of power flow based on hybrid enhanced grey wolf optimization and dragonfly algorithm | |
| Adetunji et al. | An optimization planning framework for allocating multiple distributed energy resources and electric vehicle charging stations in distribution networks | |
| Agrawal et al. | A multi-objective thermal exchange optimization model for solving optimal power flow problems in hybrid power systems | |
| CN108879691A (en) | A kind of method and device that extensive continuous tide calculates | |
| CN114492814A (en) | Method, device and medium for simulating energy of target system based on quantum computation | |
| Wang et al. | Transmission network dynamic planning based on a double deep-Q network with deep ResNet | |
| CN113094899A (en) | Random power flow calculation method and device, electronic equipment and storage medium | |
| Xue et al. | Enterprise information system structure optimization based on time property with improved immune genetic algorithm and binary tree | |
| Chen et al. | Transmission network expansion planning with wind energy integration: A stochastic programming model | |
| Chen et al. | Research on high performance computing of power system based on machine learning algorithm | |
| CN108599173B (en) | Method and device for solving batch power flows | |
| Tchorzewski et al. | Modeling and simulation inspired by quantum methods of the Polish Electricity Stock Exchange | |
| CN107423259A (en) | A kind of GPU of domino optimization accelerates trigonometric equation group back substitution method on electric power | |
| Garcia et al. | GPU-accelerated Poincaré map method for harmonic-oriented analyses of power systems | |
| Yang et al. | Enhanced multi-objective marine predator algorithm for dynamic economic-grid fluctuation dispatch with plug-in electric vehicles | |
| Ding | Power System Operation with Large Scale Stochastic Wind Power Integration: Interval Arithmetic Based Analysis and Optimization Methods | |
| Tchórzewski et al. | Quantum-inspired artificil neural networks and evolutionary algorithms methods applied to modeling of the polish electric power exchange using the day-ahead market data | |
| CN110309982A (en) | Methods of electric load forecasting, device and equipment based on matrix decomposition | |
| Li et al. | Two-stage cooperative real-time autonomy control strategy considering vulnerability of active distribution networks | |
| Tchórzewski et al. | Systemic Method of the ANN Improvement Using Evolutionary Algorithm Inspired by Quantum Solutions Based on the Electrical Power Exchange | |
| Kuzmina | CAD system for perspective power supply network development task solution | |
| Ruppert et al. | Dynamic optimal power flow with storage restrictions using augmented lagrangian algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |